大连海洋大学学报
大連海洋大學學報
대련해양대학학보
JOURNAL OF DALIAN FISHERIES UNIVERSITY
2015年
3期
324-329
,共6页
近海水质%细分外推法%模糊预测法
近海水質%細分外推法%模糊預測法
근해수질%세분외추법%모호예측법
water quality in coastal waters%subdivision extrapolation method%fuzzy forecasting method
近海水质非线性时间序列通常由于采集范围大、时间间隔长带有一定震荡性和模糊性,这使得对其进行分析与预测有一定的难度。本研究中以某近海水质指标磷酸盐( PO3-4-P)、硝酸盐( NO-3-N)、亚硝酸盐( NO-2-N)、铵盐( TNH+4-N)和硅酸盐( SiO2-3-Si)所形成的5种时间序列为例,采用逼近细分模式导出的细分外推法和多参考加权数据的模糊预测法对近海水质时序预测进行了比较分析,并通过图形与误差计算比较了两种方法的异同。结果表明:采用细分外推法预测序列在整体形状上能更好地逼近初始时序,而模糊预测法在整体逼近精度上占有优势。本研究中提出的预测比较方法可为同类问题的预测与模型选取提供参考依据。
近海水質非線性時間序列通常由于採集範圍大、時間間隔長帶有一定震盪性和模糊性,這使得對其進行分析與預測有一定的難度。本研究中以某近海水質指標燐痠鹽( PO3-4-P)、硝痠鹽( NO-3-N)、亞硝痠鹽( NO-2-N)、銨鹽( TNH+4-N)和硅痠鹽( SiO2-3-Si)所形成的5種時間序列為例,採用逼近細分模式導齣的細分外推法和多參攷加權數據的模糊預測法對近海水質時序預測進行瞭比較分析,併通過圖形與誤差計算比較瞭兩種方法的異同。結果錶明:採用細分外推法預測序列在整體形狀上能更好地逼近初始時序,而模糊預測法在整體逼近精度上佔有優勢。本研究中提齣的預測比較方法可為同類問題的預測與模型選取提供參攷依據。
근해수질비선성시간서렬통상유우채집범위대、시간간격장대유일정진탕성화모호성,저사득대기진행분석여예측유일정적난도。본연구중이모근해수질지표린산염( PO3-4-P)、초산염( NO-3-N)、아초산염( NO-2-N)、안염( TNH+4-N)화규산염( SiO2-3-Si)소형성적5충시간서렬위례,채용핍근세분모식도출적세분외추법화다삼고가권수거적모호예측법대근해수질시서예측진행료비교분석,병통과도형여오차계산비교료량충방법적이동。결과표명:채용세분외추법예측서렬재정체형상상능경호지핍근초시시서,이모호예측법재정체핍근정도상점유우세。본연구중제출적예측비교방법가위동류문제적예측여모형선취제공삼고의거。
Usually the nonlinear times series of water quality in coastal waters has uncertainty on analysis and pre-diction due to certain vibration and fuzziness caused by wide range collection and large interval. The comparison a-nalysis of the nonlinear times series of water quality including phosphate ( PO3-4 -P ) , nitrate ( NO-3-N ) , nitrite ( NO-2-N) , ammonium ( TNH+4-N) and silicate( SiO2-3 -Si) sampled in coastal waters was conducted by subdivision extrapolation, derived from the approximating subdivision scheme, and fuzzy forecasting with much weighted refer-ence data. The comparison of similarities and differences between the two methods through graphics and error calcu-lation revealed that the subdivision extrapolation showed good forecast sequence close to the initial sequence on the whole shape. The fuzzy forecasting method had the advantage of integral approximation accuracy. The findings pro-vide the reference for the solution of the similar prediction and model selection.